Update vendored DuckDB sources to 22e6d1e375

Author: duckdblabs-bot

src/duckdb/src/execution/index/art/art.cpp

@@ -4,7 +4,11 @@
 #include "duckdb/common/unordered_map.hpp"
 #include "duckdb/common/vector_operations/vector_operations.hpp"
 #include "duckdb/execution/expression_executor.hpp"
+#include "duckdb/execution/index/art/art_builder.hpp"
 #include "duckdb/execution/index/art/art_key.hpp"
+#include "duckdb/execution/index/art/art_merger.hpp"
+#include "duckdb/execution/index/art/art_operator.hpp"
+#include "duckdb/execution/index/art/art_scanner.hpp"
 #include "duckdb/execution/index/art/base_leaf.hpp"
 #include "duckdb/execution/index/art/base_node.hpp"
 #include "duckdb/execution/index/art/iterator.hpp"
@@ -21,9 +25,6 @@
 #include "duckdb/storage/metadata/metadata_reader.hpp"
 #include "duckdb/storage/table/scan_state.hpp"
 #include "duckdb/storage/table_io_manager.hpp"
-#include "duckdb/execution/index/art/art_scanner.hpp"
-#include "duckdb/execution/index/art/art_merger.hpp"
-#include "duckdb/execution/index/art/art_operator.hpp"
 
 namespace duckdb {
 
@@ -424,76 +425,17 @@ void ART::GenerateKeyVectors(ArenaAllocator &allocator, DataChunk &input, Vector
 }
 
 //===--------------------------------------------------------------------===//
-// Construct from sorted data.
+// Build from sorted data.
 //===--------------------------------------------------------------------===//
 
-bool ART::ConstructInternal(const unsafe_vector<ARTKey> &keys, const unsafe_vector<ARTKey> &row_ids, Node &node,
-                            ARTKeySection &section) {
-	D_ASSERT(section.start < keys.size());
-	D_ASSERT(section.end < keys.size());
-	D_ASSERT(section.start <= section.end);
-
-	auto &start = keys[section.start];
-	auto &end = keys[section.end];
-	D_ASSERT(start.len != 0);
-
-	// Increment the depth until we reach a leaf or find a mismatching byte.
-	auto prefix_depth = section.depth;
-	while (start.len != section.depth && start.ByteMatches(end, section.depth)) {
-		section.depth++;
-	}
-
-	if (start.len == section.depth) {
-		// We reached a leaf. All the bytes of start_key and end_key match.
-		auto row_id_count = section.end - section.start + 1;
-		if (IsUnique() && row_id_count != 1) {
-			return false;
-		}
-
-		reference<Node> ref(node);
-		auto count = UnsafeNumericCast<uint8_t>(start.len - prefix_depth);
-		Prefix::New(*this, ref, start, prefix_depth, count);
-		if (row_id_count == 1) {
-			Leaf::New(ref, row_ids[section.start].GetRowId());
-		} else {
-			// Loop and insert the row IDs.
-			// We cannot use Construct in the leaf because row IDs are not sorted.
-			ArenaAllocator arena(BufferAllocator::Get(db));
-			for (idx_t i = section.start; i < section.start + row_id_count; i++) {
-				ARTOperator::Insert(arena, *this, ref, row_ids[i], 0, row_ids[i], GateStatus::GATE_SET, nullptr,
-				                    IndexAppendMode::DEFAULT);
-			}
-			ref.get().SetGateStatus(GateStatus::GATE_SET);
-		}
-		return true;
-	}
-
-	// Create a new node and recurse.
-	unsafe_vector<ARTKeySection> children;
-	section.GetChildSections(children, keys);
-
-	// Create the prefix.
-	reference<Node> ref(node);
-	auto prefix_length = section.depth - prefix_depth;
-	Prefix::New(*this, ref, start, prefix_depth, prefix_length);
-
-	// Create the node.
-	Node::New(*this, ref, Node::GetNodeType(children.size()));
-	for (auto &child : children) {
-		Node new_child;
-		auto success = ConstructInternal(keys, row_ids, new_child, child);
-		Node::InsertChild(*this, ref, child.key_byte, new_child);
-		if (!success) {
-			return false;
-		}
-	}
-	return true;
-}
+ARTConflictType ART::Build(unsafe_vector<ARTKey> &keys, unsafe_vector<ARTKey> &row_ids, const idx_t row_count) {
+	ArenaAllocator arena(BufferAllocator::Get(db));
+	ARTBuilder builder(arena, *this, keys, row_ids);
+	builder.Init(tree, row_count - 1);
 
-bool ART::Construct(unsafe_vector<ARTKey> &keys, unsafe_vector<ARTKey> &row_ids, const idx_t row_count) {
-	ARTKeySection section(0, row_count - 1, 0, 0);
-	if (!ConstructInternal(keys, row_ids, tree, section)) {
-		return false;
+	auto result = builder.Build();
+	if (result != ARTConflictType::NO_CONFLICT) {
+		return result;
 	}
 
 #ifdef DEBUG
@@ -504,7 +446,8 @@ bool ART::Construct(unsafe_vector<ARTKey> &keys, unsafe_vector<ARTKey> &row_ids,
 	it.Scan(empty_key, NumericLimits<row_t>().Maximum(), row_ids_debug, false);
 	D_ASSERT(row_count == row_ids_debug.size());
 #endif
-	return true;
+
+	return ARTConflictType::NO_CONFLICT;
 }
 
 //===--------------------------------------------------------------------===//

src/duckdb/src/execution/index/art/art_builder.cpp

@@ -0,0 +1,91 @@
+#include "duckdb/execution/index/art/art_builder.hpp"
+
+#include "duckdb/execution/index/art/art_key.hpp"
+#include "duckdb/execution/index/art/prefix.hpp"
+#include "duckdb/execution/index/art/leaf.hpp"
+#include "duckdb/execution/index/art/art_operator.hpp"
+
+namespace duckdb {
+
+ARTConflictType ARTBuilder::Build() {
+	while (!s.empty()) {
+		// Copy the entry so we can pop it.
+		auto entry = s.top();
+		s.pop();
+
+		D_ASSERT(entry.start < keys.size());
+		D_ASSERT(entry.end < keys.size());
+		D_ASSERT(entry.start <= entry.end);
+
+		auto &start = keys[entry.start];
+		auto &end = keys[entry.end];
+		D_ASSERT(start.len != 0);
+
+		// Increment the depth until we reach a leaf or find a mismatching byte.
+		auto prefix_depth = entry.depth;
+		while (start.len != entry.depth && start.ByteMatches(end, entry.depth)) {
+			entry.depth++;
+		}
+
+		// True, if we reached a leaf: all bytes of start_key and end_key match.
+		if (start.len == entry.depth) {
+			// Get the number of row IDs in the leaf.
+			auto row_id_count = entry.end - entry.start + 1;
+			if (art.IsUnique() && row_id_count != 1) {
+				return ARTConflictType::CONSTRAINT;
+			}
+
+			reference<Node> ref(entry.node);
+			auto count = UnsafeNumericCast<uint8_t>(start.len - prefix_depth);
+			Prefix::New(art, ref, start, prefix_depth, count);
+
+			// Inline the row ID.
+			if (row_id_count == 1) {
+				Leaf::New(ref, row_ids[entry.start].GetRowId());
+				continue;
+			}
+
+			// Loop and insert the row IDs.
+			// We cannot iterate into the nested leaf with the builder
+			// because row IDs are not sorted.
+			for (idx_t i = entry.start; i < entry.start + row_id_count; i++) {
+				ARTOperator::Insert(arena, art, ref, row_ids[i], 0, row_ids[i], GateStatus::GATE_SET, nullptr,
+				                    IndexAppendMode::DEFAULT);
+			}
+			ref.get().SetGateStatus(GateStatus::GATE_SET);
+			continue;
+		}
+
+		// Create the prefix. Returns early, if the prefix_length is zero.
+		reference<Node> ref(entry.node);
+		auto prefix_length = entry.depth - prefix_depth;
+		Prefix::New(art, ref, start, prefix_depth, prefix_length);
+
+		vector<idx_t> child_offsets;
+		child_offsets.emplace_back(entry.start);
+		for (idx_t i = entry.start + 1; i <= entry.end; i++) {
+			if (keys[i - 1].data[entry.depth] != keys[i].data[entry.depth]) {
+				child_offsets.emplace_back(i);
+			}
+		}
+
+		// Create a new node containing the children.
+		Node::New(art, ref, Node::GetNodeType(child_offsets.size()));
+		auto start_offset = child_offsets[0];
+		for (idx_t i = 1; i <= child_offsets.size(); i++) {
+			auto child_byte = keys[start_offset].data[entry.depth];
+			// FIXME: Improve performance by either returning a reference to the child directly,
+			// FIXME: or by calling InsertChild after processing the child (at the end of the stack loop).
+			Node::InsertChild(art, ref, child_byte);
+			auto child = ref.get().Node::GetChildMutable(art, child_byte, true);
+			auto end_offset = i != child_offsets.size() ? child_offsets[i] - 1 : entry.end;
+			s.emplace(*child, start_offset, end_offset, entry.depth + 1);
+			start_offset = end_offset + 1;
+		}
+	}
+
+	// We exhausted the stack.
+	return ARTConflictType::NO_CONFLICT;
+}
+
+} // namespace duckdb

src/duckdb/src/execution/index/art/art_key.cpp

@@ -163,27 +163,4 @@ idx_t ARTKey::GetMismatchPos(const ARTKey &other, const idx_t start) const {
 	return DConstants::INVALID_INDEX;
 }
 
-//===--------------------------------------------------------------------===//
-// ARTKeySection
-//===--------------------------------------------------------------------===//
-
-ARTKeySection::ARTKeySection(idx_t start, idx_t end, idx_t depth, data_t byte)
-    : start(start), end(end), depth(depth), key_byte(byte) {
-}
-
-ARTKeySection::ARTKeySection(idx_t start, idx_t end, const unsafe_vector<ARTKey> &keys, const ARTKeySection &section)
-    : start(start), end(end), depth(section.depth + 1), key_byte(keys[end].data[section.depth]) {
-}
-
-void ARTKeySection::GetChildSections(unsafe_vector<ARTKeySection> &sections, const unsafe_vector<ARTKey> &keys) {
-	auto child_idx = start;
-	for (idx_t i = start + 1; i <= end; i++) {
-		if (keys[i - 1].data[depth] != keys[i].data[depth]) {
-			sections.emplace_back(child_idx, i - 1, keys, *this);
-			child_idx = i;
-		}
-	}
-	sections.emplace_back(child_idx, end, keys, *this);
-}
-
 } // namespace duckdb

src/duckdb/src/execution/index/art/node.cpp

@@ -198,31 +198,31 @@ void Node::DeleteChild(ART &art, Node &node, Node &prefix, const uint8_t byte, c
 //===--------------------------------------------------------------------===//
 
 template <class NODE>
-unsafe_optional_ptr<Node> GetChildInternal(ART &art, NODE &node, const uint8_t byte) {
+static unsafe_optional_ptr<Node> GetChildInternal(ART &art, NODE &node, const uint8_t byte, const bool unsafe) {
 	D_ASSERT(node.HasMetadata());
 
 	auto type = node.GetType();
 	switch (type) {
 	case NType::NODE_4:
-		return Node4::GetChild(Node::Ref<Node4>(art, node, type), byte);
+		return Node4::GetChild(Node::Ref<Node4>(art, node, type), byte, unsafe);
 	case NType::NODE_16:
-		return Node16::GetChild(Node::Ref<Node16>(art, node, type), byte);
+		return Node16::GetChild(Node::Ref<Node16>(art, node, type), byte, unsafe);
 	case NType::NODE_48:
-		return Node48::GetChild(Node::Ref<Node48>(art, node, type), byte);
+		return Node48::GetChild(Node::Ref<Node48>(art, node, type), byte, unsafe);
 	case NType::NODE_256: {
-		return Node256::GetChild(Node::Ref<Node256>(art, node, type), byte);
+		return Node256::GetChild(Node::Ref<Node256>(art, node, type), byte, unsafe);
 	}
 	default:
 		throw InternalException("Invalid node type for GetChildInternal: %d.", type);
 	}
 }
 
 const unsafe_optional_ptr<Node> Node::GetChild(ART &art, const uint8_t byte) const {
-	return GetChildInternal(art, *this, byte);
+	return GetChildInternal(art, *this, byte, false);
 }
 
-unsafe_optional_ptr<Node> Node::GetChildMutable(ART &art, const uint8_t byte) const {
-	return GetChildInternal(art, *this, byte);
+unsafe_optional_ptr<Node> Node::GetChildMutable(ART &art, const uint8_t byte, const bool unsafe) const {
+	return GetChildInternal(art, *this, byte, unsafe);
 }
 
 template <class NODE>

src/duckdb/src/execution/operator/schema/physical_create_art_index.cpp

@@ -114,7 +114,7 @@ SinkResultType PhysicalCreateARTIndex::SinkSorted(OperatorSinkInput &input) cons
 	auto art = make_uniq<ART>(info->index_name, l_index->GetConstraintType(), l_index->GetColumnIds(),
 	                          l_index->table_io_manager, l_index->unbound_expressions, storage.db,
 	                          l_index->Cast<ART>().allocators);
-	if (!art->Construct(l_state.keys, l_state.row_ids, l_state.key_chunk.size())) {
+	if (art->Build(l_state.keys, l_state.row_ids, l_state.key_chunk.size()) != ARTConflictType::NO_CONFLICT) {
 		throw ConstraintException("Data contains duplicates on indexed column(s)");
 	}
 

src/duckdb/src/function/table/version/pragma_version.cpp

@@ -1,5 +1,5 @@
 #ifndef DUCKDB_PATCH_VERSION
-#define DUCKDB_PATCH_VERSION "0-dev3109"
+#define DUCKDB_PATCH_VERSION "0-dev3141"
 #endif
 #ifndef DUCKDB_MINOR_VERSION
 #define DUCKDB_MINOR_VERSION 4
@@ -8,10 +8,10 @@
 #define DUCKDB_MAJOR_VERSION 1
 #endif
 #ifndef DUCKDB_VERSION
-#define DUCKDB_VERSION "v1.4.0-dev3109"
+#define DUCKDB_VERSION "v1.4.0-dev3141"
 #endif
 #ifndef DUCKDB_SOURCE_ID
-#define DUCKDB_SOURCE_ID "d229d97f40"
+#define DUCKDB_SOURCE_ID "22e6d1e375"
 #endif
 #include "duckdb/function/table/system_functions.hpp"
 #include "duckdb/main/database.hpp"

src/duckdb/src/include/duckdb/execution/index/art/art.hpp

@@ -92,8 +92,8 @@ class ART : public BoundIndex {
 	//! Drop the ART.
 	void CommitDrop(IndexLock &index_lock) override;
 
-	//! Construct an ART from a vector of sorted keys and their row IDs.
-	bool Construct(unsafe_vector<ARTKey> &keys, unsafe_vector<ARTKey> &row_ids, const idx_t row_count);
+	//! Build an ART from a vector of sorted keys and their row IDs.
+	ARTConflictType Build(unsafe_vector<ARTKey> &keys, unsafe_vector<ARTKey> &row_ids, const idx_t row_count);
 
 	//! Merge another ART into this ART. Both must be locked.
 	//! FIXME: Return ARTConflictType instead of a boolean.
@@ -138,9 +138,6 @@ class ART : public BoundIndex {
 	string GetConstraintViolationMessage(VerifyExistenceType verify_type, idx_t failed_index,
 	                                     DataChunk &input) override;
 
-	bool ConstructInternal(const unsafe_vector<ARTKey> &keys, const unsafe_vector<ARTKey> &row_ids, Node &node,
-	                       ARTKeySection &section);
-
 	void InitializeMergeUpperBounds(unsafe_vector<idx_t> &upper_bounds);
 	void InitializeMerge(Node &node, unsafe_vector<idx_t> &upper_bounds);
 

src/duckdb/src/include/duckdb/execution/index/art/art_builder.hpp

@@ -0,0 +1,55 @@
+//===----------------------------------------------------------------------===//
+//                         DuckDB
+//
+// duckdb/execution/index/art/art_builder.hpp
+//
+//
+//===----------------------------------------------------------------------===//
+
+#pragma once
+
+#include "duckdb/execution/index/art/art.hpp"
+#include "duckdb/common/stack.hpp"
+
+namespace duckdb {
+
+class ARTBuilder {
+public:
+	ARTBuilder() = delete;
+	ARTBuilder(ArenaAllocator &arena, ART &art, const unsafe_vector<ARTKey> &keys, const unsafe_vector<ARTKey> &row_ids)
+	    : arena(arena), art(art), keys(keys), row_ids(row_ids) {
+	}
+
+public:
+	//! Initialize the ART builder by passing a reference to the root node.
+	void Init(Node &node, const idx_t end) {
+		s.emplace(node, 0, end, 0);
+	}
+	//! Build the ART starting at the first entry in the stack.
+	ARTConflictType Build();
+
+private:
+	struct NodeEntry {
+		NodeEntry() = delete;
+		NodeEntry(Node &node, const idx_t start, const idx_t end, const idx_t depth)
+		    : node(node), start(start), end(end), depth(depth) {};
+
+		Node &node;
+		idx_t start;
+		idx_t end;
+		idx_t depth;
+	};
+
+	//! The arena holds any temporary memory allocated during the Build phase.
+	ArenaAllocator &arena;
+	//! The ART holding the node memory.
+	ART &art;
+	//! The keys to build the ART from.
+	const unsafe_vector<ARTKey> &keys;
+	//! The row IDs matching the keys.
+	const unsafe_vector<ARTKey> &row_ids;
+	//! The stack. While merging, NodeEntry elements are pushed onto of the stack.
+	stack<NodeEntry> s;
+};
+
+} // namespace duckdb